InstInfer: In-Storage Attention Offloading for Cost-Effective Long-Context LLM Inference

Pan, Xiurui, Li, Endian, Li, Qiao · arXiv · 2024